fix: SSE blockiert gunicorn-Worker - auf gthread umstellen

Mit 4 synchronen Workern hielt jede SSE-Verbindung dauerhaft einen
ganzen Worker besetzt. 4 offene Browser-Tabs -> alle anderen
Requests blockiert -> "Dateien laden dauert ewig".

Loesung: gthread worker-class mit 2 Workern x 16 Threads = 32
gleichzeitige Slots. Lang laufende SSE-Streams belegen nur je
einen Thread, regulaere Requests laufen unbeeintraechtigt.

nginx.example.conf: separater Location-Block fuer /api/sync/events
mit proxy_buffering off und 24h Read-Timeout, damit die Events
sofort durchkommen und die Verbindung nicht abbricht.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This commit is contained in:
Stefan Hacker 2026-04-12 10:33:02 +02:00
parent 5f905b4925
commit 3af2bc3312
2 changed files with 23 additions and 1 deletions

View File

@ -35,4 +35,9 @@ ENV UPLOAD_PATH=/app/data/files
EXPOSE 5000
CMD ["gunicorn", "--bind", "0.0.0.0:5000", "--workers", "4", "--timeout", "120", "wsgi:application"]
# gthread worker class so SSE long-poll connections don't starve regular requests.
# 2 workers x 16 threads = 32 concurrent slots; each SSE client just holds one thread.
CMD ["gunicorn", "--bind", "0.0.0.0:5000", \
"--worker-class", "gthread", "--workers", "2", "--threads", "16", \
"--timeout", "120", "--keep-alive", "65", \
"wsgi:application"]

View File

@ -24,6 +24,23 @@ server {
proxy_set_header Connection "upgrade";
}
# Server-Sent Events: Puffer aus, lange Read-Timeouts, sonst bricht die
# Live-Refresh-Verbindung nach ein paar Sekunden ab.
location /api/sync/events {
proxy_pass http://127.0.0.1:5000;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
proxy_set_header X-Forwarded-Proto $scheme;
proxy_http_version 1.1;
proxy_buffering off;
proxy_cache off;
proxy_read_timeout 24h;
proxy_send_timeout 24h;
chunked_transfer_encoding on;
}
# CalDAV/CardDAV braucht spezielle Methoden
location /dav/ {
proxy_pass http://127.0.0.1:5000;